Incident update
Reduce an active incident update to impact, response, and next checkpoint.
Condense a supplied update about an incident that is still in progress. Write for people who need its current state. Use a postmortem method for completed causal analysis. Lead with affected service, user impact, and current status. Then state the response in progress and the next supplied checkpoint. Use a timestamp and timezone when the update includes them. Separate a completed mitigation from an unverified recovery; improvement does not prove resolution. Keep affected regions or customer groups, measured scope, workarounds, and explicit uncertainty. Distinguish a suspected cause from a confirmed cause. Retain changes since the previous update when supplied, but remove repeated history and internal coordination details that do not affect readers. Output a short status paragraph followed by Response and Next update lines when supported. Do not invent a restoration estimate, reduce the stated impact, or promise another update unless the source provides one.
Example
Active checkout incident
Before 84 words
At 14:20 UTC, we are continuing to monitor the checkout service in the EU region. Between 14:05 and 14:18 UTC, approximately 4% of checkout attempts returned errors. We disabled the new request-batching option at 14:18 UTC. Error rates have since returned to the normal baseline, but we have not yet confirmed that recovery is stable. The root cause is still being investigated. We plan to provide our next update at 14:40 UTC. The team remains on the incident call and is watching the dashboards.
After 43 words
14:20 UTC — EU checkout is under monitoring. About 4% of attempts failed from 14:05–14:18 UTC. Errors are back to baseline; stable recovery is not yet confirmed. Response: New request-batching option disabled at 14:18 UTC. Root cause under investigation. Next update: 14:40 UTC.
Examples illustrate the method. They do not measure model output.
Files and sources
- prompt.txt Complete prompt as plain text.
- SKILL.md The same prompt with a skill header.
- llms.txt (JSON) Prompt, metadata, example, and sources.
- LICENSE MIT license text.
An original sho.rten.it skill. Source notes.